Subhalingam D

Identification of Hate Spreaders on Social Media (Bachelor's Thesis)

Prof. Niladri Chatterjee

We propose a novel model that uses pre-trained word embeddings for encoding the words and incorporates the sentiment scores as weights to mark the importance of the words. It then computes a weighted sum to get the tweet representation and aggregates these to obtain the user representation. The user representation is finally fed to an ML classifier. Our model achieves an accuracy of 76% on the test set and outperforms the best model in the competition.

chaii - Hindi and Tamil Question Answering

Prof. Mausam

Fine-tuned XLM-RoBERTa for multilingual Q/A using chaii-1 dataset augmented with MLQA, XQuAD & SQuAD and attained test Jaccard score of 68.72%.

Context-Sensitive Word Sense Disambiguation

Prof. Mausam

Compared non-contextual and contextual embeddings (GloVe+BiLSTM vs BERT) using WiC dataset for WSD task.

Tweet Sentiment Classifier

Prof. Mausam

Processed tweets with tweet normalization, internet slang dictionary, stemming, etc.; vectorized with TF-IDF; fed into LR.

Rule-based Written-to-Spoken Text Converter

Prof. Mausam

Built a regex-based system that accounts for chunks with abbreviations, dates. numerical quantities and inflections. Obtained test F1-score of 97.94%.

Bankruptcy Prediction

Prof. Niladri Chatterjee

Reviewed state-of-the-art bankruptcy prediction models and observed poor recall. Hypothesized class imbalance & missing values to be the reasons. Trained an ensemble model with Mean Imputation & SMOTE on Polish companies dataset and gained 10% improvement in recall.

Adaptive Network-based Fuzzy Inference System for Diabetes Prediction

Prof. Niladri Chatterjee

Trained a Takagi–Sugeno type neuro-fuzzy model in TensorFlow for diabetes prediction and obtained accuracy of 81.3%.

Movie Recommender System

Prof. Srikanta Bedathur

Used movie metadata & description of movies to retrieve most similar movies & added filters to retrieve movies of specific genre, language, year to build content-based model. Generated user-specific ratings using SVD & achieved 0.67 FCP and 0.86 RMSE with IMDb dataset (26M ratings from 270k users & 45k movies) to build collaborative model.

Dealing with Fake News on Social Media

Prof. Srikanta Bedathur

Surveyed and analysed state-of-the-art models like dEFEND, CSI, HAN, GNN to detect fake news on social media.

Document Reranking using Pseudo-Relevance Feedback

Prof. Srikanta Bedathur

Used probabilistic query expansion and relevance model based language modeling with unigram/bigram setting & Dirichlet smoothing to rerank retreived documents and improve the MRR and nDCG scores of the system.

Vector Space Model for News Articles Retrieval

Prof. Srikanta Bedathur

Implemented end-to-end retrieval system indexed with TF-IDF weights & cosine similarity-based ranking. Added prefix searching and named entity based searching (using StanfordNER) to narrow down the results of retreival. Compressed index file by encoding differences between document IDs & reduced size by half (topped class leaderboard for index size).

Web Designing & Development for SAC, IIT Delhi

Revamped the website using CSS & Javascript for better user experience and easy accessibility & retrieval of information.

Movie Review Sentiment Classifier using RNN

Leveraged RNNs with pre-trained GloVe word embeddings and LSTM layers to build a movie review classifier. Trained the model using IMDB dataset with 50k reviews after pre-processing and obtained an accuracy of about 88%.

Triangulation Topology Analysis using Graph Theory

Prof. Subodh Kumar

Generated generic Graph data structure to store triangles, points & edges for given triangulation topology of 3D shapes. Implemented traversal algorithms to get neighbours, boundary edges, count of connected components & closest components.

Priority-based Job Scheduler

Prof. Subodh Kumar

Implemented Trie, Red-Black Tree & Max-Heap to execute jobs from users for projects based on priorities & resources. Added features for fetching job status & top budget consuming users, flushing starving jobs & updating project priorities.

Analysis of Double Hashing and Separate Chaining

Prof. Subodh Kumar

Analysed asymptotic complexities and collision rate of queries in Double Hashing and Separate Chaining techniques.

Consumer-Producer Problem with Multithreading

Prof. Subodh Kumar

Sellers add the items to a Catalogue maintained as a Priority Queue, which are sorted based on the priority of sellers and then sold to Consumers.

Symbolic Differentiation

Prof. Subhashis Banerjee

Generated a Binary Tree by parsing fully parenthesised infix expression and computed its derivative by traversal. The parser was made to support a variety of functions like algebraic, trigonometric, exponential & composite functions.

Sound Detector Switch

A potential developed across Condensor Mic 's terminals due to sound is used as input to CE transistor whose output goes to a 555 timer IC in mono-stable configuration which delays the trigger signal and the output pulse from the timer goes to D-type flip flop which changes the state of the output from ON-OFF or OFF-ON.

Product Realisation by Manufacturing

Realised a Rube Goldberg machine along with a team of about 20 students, using processes like carpentry, welding, drilling.